Journal of Protein Chemistry, Vol. 11, No. 5, 1992

Structural Differences in Active Site-Labeled Thrombin Complexes with Hirudin Isoinhibitors Judith K. Rowand 1'2 and Lawrence J. Berliner 1'3

Received January 28, 1992

Hirudin, a 65 amino acid polypeptide from the medicinal leech, is the most potent thrornbin inhibitor known to date. Recently, recombinant forms have been reported, which are as effective as the isolated forms. The studies presented here demonstrate sensitive spectroscopic methods for distinguishing binding of two recombinant hirudins, HV1 and HV2-Lys 47, with active site-labeled human a-, e- and (-thrombins. Specifically, fluorosulfonylphenyl nitroxide spin labels, dansyl fluoride and p-nitrophenylanthranilate, were employed as active sitedirected covalent reporter groups. In general, the nitroxide immobilization was greater for spin-labeled thrombin complexes with HV2-Lys 47 vs. HV1. The two fluorophore moieties, dansyl and anthraniloyl, were also sensitive to differences in HV1 and HV2-Lys 47 binding, including interactions with loop 145-150 of the thrombin structure where the e- and (thrombin cleavages exist. Speculation over sequence differences between the two isoinhibitors centers on residues 24 and 47, both of which involve either a loss or gain of charge on the side chain. KEY WORDS: Hirudin; thrombin; spin labels; fluorescence; isoinhibitors.

1. INTRODUCTION 4

chemoattraction of macrophages to damaged sites in healing (Bar-Shavit el al., 1983), and activation of the anticoagulant protein C (Esmon el al., 1987). A proposed model of thrombin inactivation is via the inclusion of a-thrombin in the blood clot, where it is exposed to macrophage elastase and cathepsin G released in the clot. These proteases can selectively cleave a-thrombin at residues Alaj49A5 and Trp148, to yield e- and (-thrombins, respectively (Brezniak el aL, 1990; Brower el al., 1987). The most potent inhibitor of thrombin known to date is the protein, hirudin, from the salivary glands of medicinal leeches. The binding of hirudin to thrombin blocks both catalytic and hormonal functions (Fenton, 1989). To date, three isoinhibitors of nearly identical primary sequence have been characterized. All three isoinhibitors afe present in all medicinal leeches; the relative amounts of each vary by geographic location of the particular species. Extrinsic reporter groups have been useful in the past in elucidating subtle differences in tertiary structure and conformational changes in proteins. For

One of the most important enzymes of the hemostatic mechanism is the regulatory enzyme thrombin (EC 3.4.21.5). Thrombin activities include cleavage of fibrinogen to fibrin monomers, platelet release (Martin el al., 1975) and aggregation reactions, 1 Department of Chemistry, The Ohio State University, 120W. 18th Avenue, Columbus, Ohio 43210. 2 Present address: University of Cincinnati, Department of Pharmacology and Medicinal Chemistry, Cincinnati, Ohio 45267-0004. 3 To whom all correspondence should be addressed. 4Abbreviations used: AT-Ill, antithrombin III; NPGB, p-nitrophenyl-p-guanidino benzoate; HV2-Lys47, recombinant hirudin form 2 with lysine at position 47; Tos-Arg-OMe, tosyl-argininylmethyl ester; m-VII (m-NCO-6NH), N-[m-(fluorosulfonyl)phenyl]-4-N-(2,2,6,6-tetramethyl-piperidine-l-oxyl) urea; p-Ill (p-CO-6-OH), 3-(2,2,5,5-tetramethyl-pyrrolidine-1-oxyl)-p(fluorosulfonyl)benzoate; p-IV (p-CO-6NH), 4-(2,2,6,6-tetramethyl-piperidine-l-oxyl)-p-(fluorosulfonyl)benzamide; ESR, electron spin resonance; SDS-PAGE,. SDS polyacrylamide electrophoresis. 5 Chymotrypsinogen numbering system (Hartley and Shorten, 1971).

483 0277-8033/92/1000-0483506.50/0 © 1992PlenumPublishingCorporation

484

example, the active site-directed fluorescent inhibitor analogs, dansyl fluoride and p-nitrophenyl anthranilate, have been elegantly shown to detect subtle differences in hirudin analog binding to human and bovine a-thrombins (J. K. Rowand and L. J. Berliner, to be published). Additionally, both these labels and activesite directed phenylsulfonyl nitroxide spin-label inhibitors have also been shown to discriminate between proteolyzed human thrombin forms (Berliner el al., 1981). The labels used in this study were employed to examine the binding of two hirudin is.forms to thrombin. These studies are the first step in elucidating the function and significance of three similar is.forms in one organism. 2. EXPERIMENTAL PROCEDURES

Rowand and Berliner HV1 HV2-Lys47

Val-Val-Tyr-Thr-Asp-Cys-Thr-Glu-Ser-Gly Ile-Thr-Tyr-Thr-Asp-Cys-Thr-Glu-Ser-Gly 1 10

HV1 HV2-Lys47

Gln-Asn-Leu-Cys-Leu-Cys-Glu-Gly-Ser-Asn Gln-Asn-Leu-Cys-Leu-Cys-Glu-Gly-Ser-Asn 11 20

HV1 HV2-Lys47

Val-Cys-Gly-GIn-Gly-Asn-Lys-Cys-Ile-Leu Val-Cys-Gly-Lys-Gly-Asn-Lys-Cys-lle-Leu 21 30

HV1 HV2-Lys47

Gly-Ser-Asp-Gly-Glu-Lys-Asn-Gln-Cys-Val Gly-Ser-Asn-Gly-Lys-Gly-Asn-Gln-Cys-Val 31 40

HVI HV2-Lys47

Thr-Gly-Glu-Gly-Thr -Pro-Lys-Pro-Gln-Ser Thr-Gly-Glu-Gly-Thr-Pro-Asn-Pro-Glu-Ser 41 50

HV1 HV2-Lys47

His-Asn-Asp-Gly-Asp-Phe-Glu-Glu-Ile-Pro His-Asn-Asn-Gly-Asp-Phe-Glu-Glu-lle-Pro 51 60

HV 1 HV2

Glu-Glu-Tyr-Leu-Gln Glu-Glu-Tyr-Leu-Gln 61 65

2.1. Proteins

Electrophoretically pure a-thrombin was prepared by Echis earinatus activation of human prothrombin concentrate (gift from Dr. J. W. Fenton II, New York State Department of Health, Albany, New York), yielding a highly pure product of specific activity, 2100-2800 NIH U/rag, 89-95% NPGB active sites (Fenton el al., 1977). Thrombin (Mr 36,600) concentration was estimated by absorbance, e280= 1.73 ml mg -1 cm -~. Recombinant nonsulfated hirudins were generously donated by Dr. Hugo Grossenbacher (HV1) (CIBA-GEIGY Ltd., Basel, Switzerland) and Dr. Carolyn Roitsch (HV2Lys47) (Transgene S.A., Strasbourg, France). Hirudin concentrations were determined spectrophotometrically using ~e275= 3099 M - 1 cm- 1 (Grossenbacher, personal communication) for HV1 (Mr 6964) and e:80= 3082 M - 1 cm- 1 (Roitsch, personal communication) for HV2-Lys47 (Mr 6848). 2.2. Chemicals

The fluorescent labels, dansyl fluoride and pnitrophenyl anthranilate, were purchased from Molecular Probes (Eugene, Oregon). The fluorosulfonylphenyl nitroxide spin labels were synthesized after Wong el al., (1974). Porcine pancreatic elastase (lot 128F-8185), a-chymotrypsin agarose (lot 38F9675), phenylmethylsulfonyl fluoride (tosyl fluoride, lot 25F-0207), and Tos-Arg-OMe (lot 77C-0415) were from Sigma Chemical Co., while NPGB (lot 9009) was from ICN Pharmaceuticals. Dioxane (Fisher Scientific, Inc., lot 794789) was passed through an alumina column before use to remove trace peroxides.

Fig. 1. Primary structure of recombinant hirudins HVI and HV2Lys47. Sequence differences are highlighted in bold print.

All other chemicals were reagent grade and were used without further purification.

2.3. Methods Enzyme activity was measured spectrophotometrically with Tos-Arg-OMe (Hummel, 1959) on a Perkin-Elmer Lambda-5. SDS-PAGE (nonreducing) was carried out using a BioRad Mini Protean Cell (Laemmli, 1970). 2.4. Preparation of Modified ~- and ff-thrombin

With slight modifications, e-thrombin was prepared after the method of Brower el al. (1987). Typically, the reaction was carried out with 100-fold Table I. Summary of ESR Spectral Data (2Tu) Complexes of Spin

Labeled Thrombins with the Recombinant Hirudins, HVI and HV2-Lys47

Spin label p-lII p-IV m-VII

a-Thrombin HVI HV2

e-Thrombin HV1 HV2

ff-Thrombin HVI HV2

62.0 ~ 65.4

61.7 62.7 54.0

59.7 63.9 52.7

56.5

Accurate to 4- 0.5 G.

64.2 63.9 61.7

63.2 62.4 63.3

63.2 63.7 63.6

Thrombin Structural Differences with Hirudin Isoinhibitors

485

/

HV1

[ O ~ S02F H-N !

C=O

! H-N

J; I

o m-VII {m-NCO-6NHI

HV2-Lys47

t

2F

t Fig. 2. X-band ESR spectra of spin-labeled human s-thrombin (50-60 pM) complexed with HVI ( ~ 7 0 p M ) and HV2-Lys47(~70pM), respectively. The upper spectra are m-VII-sthrombin; the lower spectra are p-Ill-s-thrombin. All spectra were measured atpH 6.5, 0.05 M Tris, 0.75 M NaC1, 20 :E2°C. Protein concentration was typically 50-90/aM.

i•

C=O i

0 1 0

(p-CO-6OH)

HV2-Lys47

molar excess thrombin in 0.75 M NaCI, 0.05 M Tris, pH 8.0. ~'-thrombin was prepared after the method of Brezniak el al. (1990) with slight modifications. A solution of a-thrombin and a-chymotrypsin-agarose were gently stirred for 21 hr, after which the chymotrypsin agarose was removed by centrifugation. Both s- and ~'-forms were characterized on SDS-PAGE by two distinct bands at Mr 13,500 and 18,500, and the absence of the higher M,. a-thrombin band.

2.5. Fluorescence Labeling Labeling with dansyl fluoride, p-nitrophenyl anthranilate, and tosyl fluoride were after the procedures of Berliner and Shen (1977). Briefly, the labeling was carried out in 0.75 M NaCl, pH 8.0, 10% (v/v) dioxane cosolvent, for 12 hr at room temperature, followed by 4 days at 4°C. Subsequently, the

t sample was exhaustively dialyzed vs. 50 mM Tris, 0.75 M NaC1, pH 6.5. The labeling stoichiometry was typically 50-99% (dansyl) and 40-80% (anthraniloyl), as based on inhibition of esterase activity. All fluorescence spectra were measured on a Perkin-Elmer MPF-44A spectrofluorometer at 25 ± I°C.

2.6. Spin Labeling Thrombin (2 mg/ml) was labeled with a 20-fold molar excess of fluorosulfonylphenyl nitroxide in 0.05 M Tris, 0.75 M NaC1, 10% (v/v) acetonitrile, pH 7.2 for 1-2 hr, followed by exhaustive dialysis against 0.05 M Tris, 0.75 M NaCl, pH 6.5 to remove unreacted label. Labeling stoichiometries were typically 85-99% based on inhibition of Tos-Arg-OMe activity. All ESR spectra were measured on a Varian E4 spectrometer equipped with a Varian E-257 variable temperature accessory set at 20 :k 2°C. Typical

486

Rowand and Berliner

~"

SO2 F

H-N

.... ~

C-O I H-N

I 0 m-vii

HV2-Lys47 ~/

( m-NCO-6NH }

f

HVI~ o

HV2-kys47

V

o

p-lit ( p - CO-60H)

t

_j HV1

C=O I

HV2-kys47 p-IV

t

instrument settings were: microwave frequency, 9.15 GHz; microwave power, 20 mW; modulation frequency, 100MHz; applied field, 3255 G; scan range, 100G; modulation amplitude, 1.25 G; scan time, 16 rain; and time constant, 0.250 sec. The hyperfine extrema, 2Tll, were measured at three- to fivefold higher gain and two- to fourfold higher modulation amplitude to enhance these peaks.

3. RESULTS Figure 1 lists the primary sequences of the two recombinant hirudin forms, HVI and HV2-Lys47. These differ by eight residues, which are indicated in bold.

(p-CO-6NHI

Fig. 3. X-band ESR spectra of spin-labeled human a-thrombin (50-60/zM) comptexed with HV1 (~70/zM) and HV2-Lys47 (70 pM), respectively. The upper spectra are on m-Vll-a-thrombin, the middle spectra are p-lll-a-thrombin and the lower spectra are p-lV-a-thrombin, the middle spectra are p-IIl-a-thrombin, and the lower spectra are p-IV-(-thrombin. All other conditions are as in Fig. 2.

Figure 2 depicts ESR spectra for HV1 and HV2Lys47, respectively, complexed with m-VII- and p-IlIe-labeled-thrombins. The differences in immobilization of the nitroxide in the active site region are quite obvious between HV2-Lys47 and HV1 when comparing the hyperfine extrema 2TI] (indicated byarrows in Fig. 2). The larger splitting of HV2-Lys47 indicates a greater nitroxide immobilization vs. the HVl-complex. Addition of the two hirudin isoinhibitors to mVII and p-III-labeled-(-thrombins also showed a similar discrimination with more nitroxide immobilization in the HV2-Lys47 vs. HV1- complex (data not shown). Figure 3 depicts ESR spectra for m-VII and p-III-labeled a-thrombin, which showed similar trends to those derivatives above. However, p-IV-athrombin showed the opposite behavior for the HV1 complex which gave the larger immobilization. On the

Thrombin Structural Differences with Hirudin lsoinhibitors

other hand, p-IV-e- and p-IV-(-thrombin exhibited identical spittings (within experimental error) in the presence of both hirudin isoinhibitors. The results are summarized in Table I. Figure 4 depicts the extrinsic fluorescence emission spectra of anthraniloyl-~'-thrombin in the presence of HV1 and HV2-Lys47. Here the HVl complex shows a slightly smaller quantum yield. Table II tabulates the changes of fluorescence intensity observed

90

487

upon addition of the hirudins to various labeled thrombins. Both dansyl- and anthraniloyl-thrombins displayed a greater quantum yield in complex with HV2-Lys47 vs. HV1. The only exceptions were the dansyl-~'-thrombin complexes, where the relative fluorescence intensities were equal, as was also the case for the intrinsic fluorescence emission intensity of a-thrombin complexes. On the other hand, a serine 195 blocked enzyme (i.e., tosyl-a-thrombin) yielded different intrinsic fluorescence quantum yields, resulting in a larger increase for the HV2-Lys47 complex vs. the HV1 complex. Except for the 2 nm blue shift in Z ~ ~ for anthraniloyl-~'-thrombin upon complexation with either hirudin isoinhibitor, no other fluorescence bands were shifted (Table I).

4. DISCUSSION . -0-3 "1CO (D (&.

45

>

rr

0 460

380 Wavelength (nm)

Fig. 4. Fluorescence extrinsic emission spectra of 2.00/IM anthraniloyl-f-thrombin in the presence of 2.02/tM HV1 and 2.04/~M HV2-Lys47. Conditions were ,Lx=350nm, pH6.5, 0.75 M NaCI, 0.05 M Tris, 25+ I°C. Table II. PercentChangeof FluorescenceEmissionIntensityupon

Complexationof VariousThrombinDerivativeswithRecombinant Hirudins, HVI and HV2-Lys47 % Changeextrinsicemission intensity Label None a Tosyl- a -thrombin" Dansyl- e-thrombin Dansyl-~'-thrombin Anthranfloyl-a-thrombin Anthraniloyl- e-thrombin Anthraniloyl-~-thrombin

HV1 25 15 - 10 - 10 20 NC b NC b

"Intrinsic fluorescencewas monitored. bNC, No change.

HV2-Lys47 25 20 NC b - 10 35 5 15

The biological advantage in having three similar yet distinct peptides to perform the same function still remains a mystery (Johnson el al., 1989). The answer may be difficult to describe, even with X-ray structures of several thrombin complexes. It is obviously important to understand the differences between the peptides in terms of interactions with thrombin at the molecular level. The data presented in this paper strongly suggest that structural differences do exist between the HV1 and HV2-Lys47-thrombin complexes. These are monitored in the active site region but may also be conformationally elicited at the catalytic center from a more distant binding locus. The larger 2TIi observed for HV2-Lys47 : spin-labeled thrombin complexes in the majority of our ESR results suggests that the binding interactions with HV2-Lys47 result in a greater nitroxide immobilization than with HV1 (a difference of _>1.0 gauss is typically considered to be significant). Since the nitroxide moieties of these labels (e.g., p-III and m-VII) are known to occupy spalially distinct regions in the extended active site of thrombin (Berliner el aL, 1981), the differences observed above may be localized to the phenylsulfonyl binding locus rather than the nitroxide group (which extends approximately 12-15 A from the catalytic Ser195). Another possibility is that HV2-Lys47 more completely "blankets" the active site upon binding, resulting in less rotational space for the nitroxide moiety. The spin label p-IV is isostructural to p-III, differing only by an amide vs. an ester linkage between the nitroxide moiety and the phenyl ring, respectively. The 2Tii for p-IV-thrombin when complexed with either hirudin isoinhibitor was typically 63 G, which

488

was also identical to that observed for HV2-Lys47: pllI-e-or-~'-tbrombin complexes. However, the 2Tii for HVI: p-III-thrombin complexes was always smaller, ranging from 59.7-62.0 G. One possible contribution to this difference is from the amido-NH of p-IV, which may be able to hydrogen bond with the inhibitor, resulting in more immobilization of the nitroxide ring. A difference in primary sequence between the two isoinhibitors might allow HV2-Lys47 to act as a hydrogen donor in this region. In addition, differences were also apparent from the fluorescence results, which indicated that these variations may exist within the catalytic site. These differences were further amplified with the fluorescent labeled e- and ~'-thrombins. The results would suggest that loop 145-150, which contains both the e- and ~'thrombin cleavage points, may be interacting differently (either directly or indirectly) with the two hirudin isoinhibitors (Brower el al., 1987; Brezniak el al., 1990). 6 Additional insight is provided from examination of their primary structures. Focusing first on the Nterminus of hirudin, we note that the crystallographic structure of HV2-Lys47 : human a-thrombin complex showed that Ile 1 was involved in nonpolar interactions at the catalytic site (Rydel el al., 1990). In HV1, the substitution of a Val for this residue should be inconsequential. Residue 2 binds at the edge of the specificity pocket and does not appear to contribute much to the overall binding. Moving toward the Cterminus, we note that both crystallographic and N M R studies showed that residues 31-36 were disordered; thus, even though the differences at residues 33, 35, and 36 are not conservative, the structural evidence suggests that they are unlikely to have an effect (Clore el al., 1987; Rydel el al., 1990). Residue 53 was also found to have poorly defined electron density in the crystal, indicating few, if any, interactions between this residue and the thrombin structure. There are, however, two residues which appear to be potentially critical in differences in thrombin binding with HVI and HV2-Lys47. In HV2-Lys47, Lys-24 is involved in a hydrogen bond to thrombin through a water molecule. The substitution Gin24 6 Due to insertions, both loops 145-150 and 70-80 contain 11 residues, while loop 59 61 contains 12.

Rowand and Berliner

should alter the structure and chemistry sufficiently to disrupt this interaction. The second residue, Glu-49, in HV2-Lys47, forms an ion pair with His-5! of thrombin. In HV1, the corresponding Gin49 invalidates such an ion pair. In summary, the two most likely hirudin residues responsible for these differences are 24 and 49. While residue 49 lies near the catalytic site, residue 24 resides nearer to thrombin loops 59-61 and 70-80. Also, these regions of the thrombin structure may be effected differently in the various structural probes outlined in these experiments. Furthermore, these results attest to the power of active site spin labels and fluorescent probes as subtle monitors of thrombin interactions. ACKNOWLEDGMENTS

This work was supported in part by a grant from the USPHS (HL24549). REFERENCES Bar-Shavit, R., Kahn, A., Fenton, J. W., II, and Wilner, G. D. (1983). Lab. Invest. 49, 702-707. Berliner, L. J., and Shen, Y. Y. L. (1977). Thromb. Res. 12, 15-25. Berliner, L. J., Bauer, R. S., Chang, T. L., Fenton, J. W., II, and Shen, Y. Y. L. (1981). Bioehemistty 20, 1831-1837. Brezniak, D. V., Brower, M. S., Witting, J. I., Walz, D. A,, and Fenton, J. W., II (1990). Biochemistry 29, 3536-3542. Brower, M. S., Walz, D. A., Garry, K. E., and Fenton, J. W., II (1987). Blood 69, 813-819. Clore, G. M., Sukumaran, D. K., Nilges, M., Zarbock, J., and Gronenborn, A. M. (1987). EMBO J. 6, 529-537. Esmon, C. T. (1987). Seience 235, 1348-1352. Fenton, J. W., II (1989). Semin. Thromb. Hemostasis 15, 265-268. Fenton, L W., II, Fasco, M. J., Stackrow, A. B., Aronson, D. L., Young, A. M., and Finlayson, J. S. (1977). J. Biol. Chem. 252, 3587-3598. Hartley, B. S., and Shotton, D. M. (1971). The Enzymes, Vol. III, Academic Press, New York, pp. 323-373. Harvey, R. P., Degryse, E., Stefani, L., Schamber, F., Cazenave, J.-P., Courtney, M., Tolstoshev, P., and Lecocq, J.-P. (1986). P~vc. Natl. Acad. Sci. 83, 1084-1088. Hummel, B. C. W. (1959). Can. J. Biochem. Physiol. 37, 13931399. Johnson, P. H., Sze, P., Winant, R., Payne, P. W., and Lazar, J. B. (1989). Semin. Thromb. Hemostasis 15, 302-315. Laernmli, U. K. (1970). Nature 227, 680-685. Markwardt, M. (1989). Semin. Thromb. Hemostasis 15, 269-282. Martin, B. M., Feinman, R. D., and Detwiler, T. C. (1975). Biochemistry 14, 1308-1314. Rydel, T. J., Ravichandran, K. G., Tulkinsky, A., Bode, W., Huber, R., Roitsch, C., and Fenton, J. W., II (1990). Science 249, 277-280. Wong, S. S., Quiggle, K., Triplett, C., and Berliner, L. J. (1974). J. Biol. Chem. 249, 1678-1682~